
@bootstrap_yangi
iAccount based inEast Asia
About this account
- Account based in
- East Asia
- Connected via
- East Asia App Store
Account-level information from X, not a live location or the device used for a specific post.
Joined March 2012
- Tweets6.9K
- Following996
- Followers190
- Likes7.7K
GSV upon further reflection... retweeted
Chinese AI cos historically have invested less in this type of thing so this is a notable development.
ByteDance has set up a dedicated frontier AI model alignment and preparedness department, Chinese media The Paper reports.
The department will work on model/agent alignment and frontier risk mitigation including CBRN (one of the first Chinese companies to say publicly that they are working on this).
Worth noting that ByteDance has always had a team working on model alignment, led by Yan Lin (严林). What's new is that this work is now being elevated to the level of a standalone department, while still being led by Yan.
thepaper.cn/newsDetail_forwa…
👀
Fantastic to see @nvidia supporting UK academic labs and the @AISecurityInst with $250m to advance technical AI safety work!
nvidianews.nvidia.com/news/n…
Is there a prediction market on when the pr would flip again and towards what direction
I think Vitalik talked about this once it's so insane!!!
The Han people literally could've bootstrapped the singularity in 700AD if it stayed fragmented to kept the intellectual innovative energy, and today ppl are still infected with the Chinese-unity-mind-virus
With recent discussions of EA influence, it’s worth taking a look at the first time an effectiveness-oriented altruistic movement held significant sway over a country: 4th century BC China.
The Mohists argued for “impartial care” as the moral standard, prioritized prevention of war as a top cause area, had members take high-paying positions in governments (i.e. the war-making bodies) to try accelerate defense over offense (and sometimes donate gold to the movement), and scorned spending on rituals and ceremonies on grounds that the poor should be given more food instead. The Daoists attacked them for demandingness and the Confucians for treating family like strangers. All lost to the Legalists when the Qin dynasty started.
When impartial welfare philosophy came to the West with Bentham, it made a big advance on the Mohists. Mohist doctrine had been afraid of ghosts. Bentham disbelieved ghosts on the grounds that no one ever saw a naked ghost. He was deathly scared of goblins instead.
Predicted this shit pre Covid; such a perfect diplomatic/intel tool and is a tilted win win for both govs
Is this hypothetical cocktail so different from the many deeply imperfect but load bearing drugs we’ve been administering for decades?
In the same way that medicine isn’t ml (and ai bros should usually sit their ass down)…
GSV upon further reflection... retweeted
Why care? Distillation (or OPD) is used in frontier model training. Our results suggest that some kinds of misalignment could propagate via distillation through unrelated data. Reward seeking, scheming and secret loyalties involve conditional policies (like backdoors) and sophisticated capabilities (e.g. inferring how rewards are calculated for an environment).
Caveat: Our experiments are in toy settings (because this makes transfer easy to measure) and future work could explore realistic post-training.
I’m totally ok with if the hillclimbing approach to agi alignment totally worked out and my hand wringing is but one of the signs of my neuroticism; but for now I really want to know how to improve the states more robustly through technical means…
The inverse p zombi: absolutely useless and a pointless existence in every respect other than being conscious (me myself)















